INTRODUCTION: Physical fitness in soccer needs to be constantly updated due to current demands in contemporary soccer. OBJECTIVE: To assess the sensitivity of the K Means Clustering in group distribution based on physiological responses relevant to soccer. METHODS: The athletes underwent anthropometric evaluations to determine fat percentage (%F) lean mass (LM), treadmill incremental test to obtain the VO2 maximum (VO2max) and ventilatory threshold velocity (VL), as well as a field test for agility (AG) and vertical jump (VJ). Data were analyzed by Kruskal-Wallis and distribution of groups was determined by K Means Clustering according to their similarities with these physiological variables, assuming significance level of p < 0.05. RESULTS: Showed that both groups were significantly different only concerning VJ (p < 0.001); LM (p < 0.001); VL (p = 0.011) and VO2max (p = 0.029) indicating that the athletes need to be distributed in groups for these variables. Nevertheless, %F and AG (p = 0.317; p = 0.922) respectively, were not different, indicating that these variables can be improved in all athletes collectively. CONCLUSION: The results make us conclude that group distribution by K Means Clustering technique can be performed using physiological responses of athletes in an attempt to optimize training for professional soccer players with focus on the common main training needs regardless of their tactical function played on the field.
cluster analysis; physical education and training; soccer
